A human-machine collaborative autonomous navigation system and method based on multi-constraint optimization
By combining Bayesian neural networks and deep reinforcement learning, a human-machine collaborative autonomous navigation system with multi-constraint optimization is developed. This system addresses the lack of safety in existing navigation methods under sudden anomalies and dynamic environments, achieving highly safe and robust navigation performance.
Patent Information
- Application Number
- CN202610856579.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-06-15
- Publication Date
- 2026-07-17
AI Technical Summary
Existing navigation methods based on deep reinforcement learning cannot adjust their decisions in a timely manner when faced with sudden abnormal situations, resulting in low decision security. Furthermore, they lack multi-objective overall optimization through human-machine collaboration and have poor adaptability in dynamic environments.
A human-machine collaborative autonomous navigation system based on multi-constraint optimization is adopted, including a navigation action generation module, a human interaction module, a confidence assessment module, a human-machine collaborative optimization module, and a Tube MPC correction module. The system uses Bayesian neural networks and deep reinforcement learning networks to perform safety confidence assessment and action fusion optimization. By combining linear velocity, angular velocity, and safety velocity constraints, a robust invariant set is constructed to ensure navigation safety.
It achieves navigation with high safety and robustness in complex dynamic environments, and ensures the safety and effectiveness of the navigation process through human-machine collaborative optimization and uncertainty-driven closed-loop control.
Smart Images
Figure CN122408735A_ABST